i'm on a roll... the kgsshv

[justin]

got it from onlinecomponents

since their communication is literally non-existent, I was hesitant to buy from them with a listed stock of 0. however, about a month ago i noticed "on order: 1" so I snatched it, figuring I could always have the credit card company cancel the payment (the price almost seemed like a scam - though it was $720 when I bought it, and now it's listed at $570). low and behold, it was real

you OWE me for that drop in price. i have 20 on order.

edit: ah. you got in before the price drop once i saw that price drop i ordered a dozen more. will probably go back for more soon. can't even fit 1 order on a credit card..

Edited June 26, 2011 by justin

[El_Doug]

you OWE me for that drop in price. i have 20 on order.

edit: ah. you got in before the price drop once i saw that price drop i ordered a dozen more. will probably go back for more soon. can't even fit 1 order on a credit card..

you could have warned me, you bastard

[justin]

at least you only ordered 1. i had an incident with 2 in a box...loose. brass on brass battle all the way from Florida. they were returned, maybe someone was fired over that one.

[kevin gilmore]

fixed duplicates on the schematic. also a few other things that

got to the board and were not on the schematic.

pathetic parts list here.

QTY PART-REFS VALUE

--- --------- -----

Resistors

---------

4 R1,R9,R21,R22 300k

4 R2,R3,R13,R14 15/3W

2 R4,R17 50k

4 R5,R15,R29,R30 440k

2 R6,R16 20k

4 R7,R10,R18,R20 240k /3W

2 R8,R19 650

4 R11,R12,R23,R24 500k

2 R25,R26 2K

1 R27 787k high voltage resistor subject to adjustment depending on transformer

1 R28 15k high voltage resistor subject to adjustment depending on transformer

1 R39 3000

2 R40,R42 100k

1 R43 5M

Capacitors

----------

4 C1,C4,C5,C9 680uF

6 C2,C3,C6,C8,C12,C14 100uF

2 C7,C17,C19,C21 .1uf /1000v

2 C10,C15 .22uf /1000v mouser 5989-1KV0.22-F

2 C11,C13 4700uF

2 C16,C18 1000pf /1000v

Integrated Circuits

-------------------

1 U3 7815

1 U4 7915

Transistors

-----------

4 Q1,Q2,Q11,Q12 fqpf8n80c

4 Q3,Q4,Q15,Q16 2sc3840

6 Q5-Q7,Q17-Q19 2sa1486

2 Q8,Q20,Q2 10m90s

4 Q9,Q10,Q13,Q14 2n3904

Diodes

------

2 D1,D6 24V zener

6 D2,D10,D12,D13, 1n4007

D21,D22

4 D3,D4,D7,D8 200V zener

2 D5,D9 100V zener

3 D17,D18,D20 150v zener

1 D19 130v zener

Miscellaneous

-------------

3 BR1-BR3 BRIDGE which is actually 8 individual diodes and one bridge

2 REF,REF2 lt1021-10

high voltage resistors are mouser hvr3700007153fr500

http://www.vishay.com/docs/30260/hvr25.pdf

ALL HEATSINKS ARE GROUNDED. The ixys parts must be mounted on the ceramic insulators

and secured with a non-conductive screw. Others that have built the previous version

had this problem, then lots of parts blow up.

You can trim this power supply for minimum heat.

For the 450 volt version, the zener string should be 500 volts

and the unreg can be as low as 550 volts. Then it will work at 10% low line.

If you want something that works 15% low line, bump the unreg to 600 volts.

Since its a crap shoot on getting transformer manufacturers to actually meet

specs you give them, its up to you to pick the appropriate unreg high voltage AC.

But something like 420 vac under load works.

You can easily change a few parts and go to 500 volt outputs.

Bump the zener string to 550 volts, and unreg to 600 volts.

at which point you don't need the voltage multiplier and can

jumper around it for the bias.

Edited June 26, 2011 by kevin gilmore

[El_Doug]

In your estimation, how much more tweaking before have a version worthy of a pcb run that truly satisfies the "any intermediate builder will have no trouble" mantra?

I wish I could help by prototyping all these things, but I fall into the "i know how to solder and use a multimeter only" club

[kevin gilmore]

This is about as bulletproof as i can make it.

With the current limit on the power supply you should

be able to do all sorts of evil things to it without

blowing up any parts.

[wink]

With the current limit on the power supply you should

be able to do all sorts of evil things to it without

blowing up any parts.

[Horio]

This is about as bulletproof as i can make it.

With the current limit on the power supply you should

be able to do all sorts of evil things to it without

blowing up any parts.

Sounds like a perfect build for me!

[kevin gilmore]

ujamerstand says that the amp board won't fit in the 3U pessante case

so i shortened the board to 3.9 vertical inches.

http://gilmore.chem.northwestern.edu/kgsshvampv7s.zip

someone needs to check this.

anyone with the real case to measure between the bracket mounting

rails?

Edited June 27, 2011 by kevin gilmore

[ujamerstand]

There's CAD drawing on hifi2000 website. http://www.hifi2000.it/redirect.asp?id=7&type=dw

Attached new drawings for the sink taps. The board now fit. http://www.sendspace.com/file/0ild9y The botom mounting positions still overlaps with the bottom bracket slightly, but it's no big deal. The bracket could be filed slightly in those positions or the board could be secured on top center and sides only.

[kevin gilmore]

moved mounting holes .05

everything should fit now.

[ujamerstand]

Confirmed Everything fits. Thanks!

[kevin gilmore]

side mount amp board for the 3U pesante case

http://gilmore.chem.northwestern.edu/kgsshvampv7s.jpg

[MASantos]

Regarding mounting the amp boards on the pesante 3G case, do the heatsinks have enough depth to allow blind tapered holes?

This board layout somehow kills the "easy to build" feature.

[nattonrice]

They do have enough base depth.

[kevin gilmore]

You can always do the amp boards with the on board heatsinks.

I'm just not willing to do another 700+ lbs of custom aluminum

extrusions.

I'm looking for a better solution, like a chassis manufacturer

that will drill and tap the heatsinks. Similar to what

front panel express does.

Except for the blind tapping part, anyone that does a careful

layout by printing a 1:1 hole map taping it to the heatsink

and then center punching it should be ok, as the parts have

plenty of mounting room. I would drill thru a total of .125

inch extra, which makes using regular taps work well.

Assuming the base of the heatsink is .250 thick, you really

want to drill and tap more than .125 anyway.

[luvdunhill]

So all the output devices are within 1 inch of the top of the extrusion??

[kevin gilmore]

the center holes of the transistors are .875 from the top of the extrusion.

This allows for .5 inch for the bracket, then .125 inch of space between

the bracket and the edge of the transistor.

Some people insist that heat travels up. Which it does a bit, but not

as much as you might think. So what you should probably do is mount

the amp boards upside down, and then the power transistors are .875 from

the bottom of the heatsink.

To make it fit into the 3.8 inch space between the 2 brackets, the

input had to be moved to the side. So its no longer an exact mirror

image for the inputs although the outputs are mirrored.

[ujamerstand]

yes. 3/4" from the top, in fact.

[luvdunhill]

I can say from personal experience that you get increased effective thermal resistance with heat sources that close to a free air edge. Don't know how it works, but it clearly does. Yes, the bottom orientation is better, but either way you wont make full use of that extrusion.

[kevin gilmore]

Then you have to use the angle brackets and build it T2 style.

But then you can't get it all in one standard box.

I'm open to suggestions.

[FrankCooter]

Very elegant power supply! Even I can now understand the "battery" sub-circuit. I've signed up for a couple of board sets, but I'm also interested in building a perf-board prototype power supply for the Egmont based amp I recently buit. Kerry, if it's no trouble, I'd really appreciate a photo of the underside of your power supply prototype. You're far neater than I'd ever be on my own.

If everything works well, I'd like to try a version with 600V rails and use it to drive an amp with 801a DHT outputs. Besides zener and resistor changes, and maybe a 1KV rated pass device, anyone see any problems?

I'm planning on building the version with the virtual battery, but appreciate the option of the simple zener string. My knowledge of this sort of thing is limited, but it looks to me if you eliminate the battery circuit, you've eliminated most of the error correction abiilty. Basically you have a stabilized but not regulated supply. If the battery cicuit proves troublesome, why not a simple error correction circuit based on a single 2SA1968 eq.?

Thanks to all that have worked on this project, especially Dr. Gilmore. It's been an education just watching this take shape!

[kevin gilmore]

You should be able to boost the thing to do 600 volts.

Primary and secondary caps in series are good for at least 800 volts

but 900 volts unreg is possible if you make sure the transformer has

a center tap to avoid imbalances in the main caps (something mikhail

never understood/or choose to ignore)

The ixys current sources are rated to 900 volts.

you can use the c4686a's to replace the 600 volt npn's, they are good to over 1kv

you would have to replace the pnp's with 2sa1968's which are good to 900 volts

there is a version of the pass fet rated at 900 volts, in fact there are ixys

parts rated at over 1200v (not cheap)

[Kerry]

Here's a couple of shots of the top and bottom. I really didn't do much in terms of optimizing the layout. I just kind of followed the schematic.

It's not the prettiest wiring on the back. I used 18 gage wire for the ground along the bottom. Also, I used 600V Teflon wire when I was going over other wires and tried to raise them off the board a bit.

EDIT: Remember, my resistor values are a bit different because I was shooting for a 425V supply.

Edited June 29, 2011 by Kerry

[luvdunhill]

Then you have to use the angle brackets and build it T2 style.

But then you can't get it all in one standard box.

I'm open to suggestions.

That or a separate board like Justin did, but that makes less sense for a solid state amp. I'm not sure what to recommend, other than try it. I've looked long at a lot of the device clip mounting methods (which still requires a 1:1 ratio of holes to devices) but you could look into a single long bar that applies pressure to all of the devices at once. You could even mill detents into the bar to align the devices. Again, doesn't solve the location issue, but might make it easier to build.